Object Identification System Using Positional Deviation Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object identification systems face inefficiencies in narrowing down the number of patterns to be collated with a target object's pattern, particularly due to variations in image position and noise patterns on object surfaces.
Innovation Solution
An object identification system that utilizes positional deviation vectors and noise patterns in images formed on object surfaces to reduce the number of patterns that need to be collated, combining these features for enhanced efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all patterns in the database are collated with the identification target pattern, then the identification accuracy is maintained, but the collation time and computational load increase significantly
Solution Approach 1:
The patent segments the pattern matching process into two stages: first extracting feature quantities (such as positional deviation vectors and noise patterns) from both the identification target and database patterns, then performing collation based on these features rather than raw patterns. This segmentation reduces the collation scope while preserving identification accuracy.
Solution Approach 2:
The patent performs preliminary extraction of feature quantities (positional deviation vectors, noise patterns) from patterns before the actual collation process. By pre-processing and storing these features, the system avoids computationally intensive operations during the collation phase, significantly reducing collation time while maintaining accuracy.
2Reliability
If the collation process considers all possible patterns, then no valid patterns are missed, but the computational complexity and processing load increase
Solution Approach 1:
The patent changes the parameters used for collation from raw pattern data to extracted feature quantities (positional deviation vectors, noise patterns). This parameter transformation simplifies the collation process by working with condensed numerical features rather than complex pattern data, reducing computational complexity while maintaining matching completeness.
Solution Approach 2:
The patent extracts essential features (positional deviation vectors and noise patterns) from complete patterns, separating the critical identification elements from the full pattern data. This extraction allows collation to focus on the most discriminative features, reducing complexity while preserving matching reliability.
3Ease of operation
If the system uses only positional information for narrowing down patterns, then the narrowing process is simple, but the reduction effectiveness is limited
Solution Approach 1:
The patent merges multiple feature types (positional deviation vectors and noise patterns) into a combined collation process. By combining these different feature dimensions, the system achieves more effective pattern narrowing than using either feature type alone, significantly improving productivity while maintaining operational simplicity through automated feature integration.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
An object identification system 1 acquires a captured image of each of a plurality of objects respectively having peculiar patterns on surfaces thereof, images being formed on the surfaces, classifies, for each of the plurality of objects, the object into any one of a plurality of classifications according to features based on the predetermined image formed on the object in the captured image acquired concerning the object, and registers, for each of the plurality of objects, identification information of the object, a pattern image indicating the pattern on the surface of the object, and a classification of the object in association with one another. The object identification system 1 acquires candidates of classifications of one target object among the plurality of objects on the basis of a classification result based on a target object captured image obtained by capturing an image of the target object, acquires pattern images registered in association with the acquired classifications, and specifies, as identification information of the target object, identification information associated with a pattern image matching the target object captured image among the acquired pattern images.